Under review, July 2001
Towards a Model of Technology and Literacy Development:
Story Listening Systems
JUSTINE CASSELL
[email protected]
MIT Media Laboratory, E15-315, 20 Ames Street, Cambridge, MA 02139, USA, (617)253-4899
Abstract: This article lays out a program of research designed to address one specific need of young children – to
learn how to write – based on one specific ability of young children – the ability to tell stories. I outline a model that
accounts for how non-screen-and-keyboard based technologies that listen to children can be used in such a way as to
support their emergent literacy behaviors and have an effect on their subsequent writing skills. The model comprises
four components: the importance of emergent literacy behaviors, which are features of literate language that are
demonstrated in children’s oral language, the critical role played by a socially-situated peer, the design of nonkeyboard-based computational technologies, and the potential of information technologies that encourage
construction rather than consumption. I describe one kind of technology that fits the model -- the story listening
system (SLS) – and describe a number of implemented story listening systems and evaluation of their use by children.
“Narrative ability is the single most important language
ability for success in school” (Feagans and Appelbaum, 1986: 359)
1. Introduction
The discussion of the role of computational technology in children’s development has become
increasingly polarized over the last year or so. On the one hand we find a frantic push to place computers
and internet access into all U.S. schools (Committee on Information Technology Literacy, 1999), and on
the other hand, a frantic push-back to place a “moratorium” (Cordes, 2000) on children’s access to
computers. Clearly the answer lies at neither end of this long spectrum, and a careful review of existent
studies shows a number of benefits, a palmful of harmful effects, and a plethora of unknowns (Wartella,
O'Keefe et al., 2000). Our own response is that the answer lies in responsible and developmentally
informed design and evaluation of technology that specifically targets the needs and unique abilities of
young children. Computers will not go away, nor should they, as they provide benefits to children that
are either unavailable elsewhere, or too resource-intensive otherwise to be practical. History also teaches
us that whereas initial fears about movies or radio or television concern the very presence of the medium,
it is the content of the medium – the message – that must be evaluated and improved (Bickham, Wright et
al., 2001; Wartella and Jennings, 2001). “Computer technology” need not be incompatible with playbased learning, physical activity, active engagement, and social interaction, those features of childhood
whose loss computer phobists decry.
This article describes a program of research designed to address one specific need of young children – to
learn how to write – based on one specific ability of young children – the ability to tell stories. I lay out a
model of how to support children learning how to write through the use of story listening systems (SLS).
(SLS). The model depends on four components: the importance of emergent literacy behaviors, which
are features of literate language that are demonstrated in children’s oral language, the critical role played
by a socially-situated peer, the design of non-keyboard-based computational technologies, and the
potential of information technologies that encourage construction rather than consumption.
Story Listening Systems - 2 -
While multiple, and technological, literacies have become the topic of considerable research, reading and
writing literacy remain the basis of education, and the prerequisites to science, mathematics, and
technology fluency. Effective writing skills are important in all stages of life, from early education to
future employment. Nevertheless, many children in the United States lack even basic writing skills. In
its 1998 Condition of Education, the United States Department of Education’s National Assessment of
Educational Progress found that only 31% of 11th graders (17 year olds) were generally able to write
complete, sufficient responses to questions, while only 2% were capable of providing effective, coherent
responses (Snyder and Wirt, 1998). Reading and writing skills are not learned only in school, however.
Children prepare themselves for later literacy in many important ways long before first grade (Heath,
1983). And preparation for writing literacy consists of more than knowing how to form letters with a
pencil. It includes: (1) treating language as an object (metalinguistic awareness (Cazden, 1976)); (2)
maintaining cohesion and reference in oral language (to introduce a new character by saying “The man
wearing blue” as opposed to “he” . . . but to refer back later by saying “he”); and (3) making one’s
communicative intentions known (for example, “I’m going to tell you about my day”).
Many of these emergent literacy (Teale and Sulzby, 1986) skills are first acquired in language play and
in storytelling. Many of them are acquired in the context of children’s interactions with peers, in early
play contexts. Peers push one another to make communicative intentions known clearly (Goncu, 1993)
and play a critical role in the acquisition of storytelling skills. For example, Preece (1992) found that
children’s interaction with each other contributed to the modification, expansion, increased coherence,
and complexity of their anecdotes and stories. Preece’s study revealed that children are active, alert,
engaged, and even aggressive listeners. Educators have begun to recognize the importance of the peer
context, and address it through classroom activities such as sharing time (Gee, 1986; Michaels, 1986;
Gallas, 1992), round-robin storytelling, and the editor’s corner.
Parents also play a role in emergent literacy activities, scaffolding early reading activities by asking
questions, expanding children’s utterances, and focusing on books with children (Whitehurst, Falco et al.,
1988). Only certain cultural contexts (primarily white mainstream/middle class families), however,
privilege early literacy connections (Heath, 1986). And teachers often have difficulty accommodating
the discourse styles of children whose backgrounds are different from their own (Michaels, 1986). In
view of this problem, programs have been started to train parents to support early pre-literacy activities,
and these programs have demonstrated successful results (Hinkle, 2000; Singer, 2001), including gains in
vocabulary acquisition and the use of decontextualized language when parents are trained to elicit
children’s narratives. But programs such as these are resource-intensive. In addition, in part because we
do not yet know exactly what activities lead to literacy, intervention programs are not as effective as they
might be. Knowledge about the bridge between emergent and actual literacy is still partial, particularly
where writing literacy is concerned. What features of oral language are precursors to what features of
written language? In what contexts are these features of oral literacy found? And, how do we design
interventions to lead to better writing literacy?
I believe that technology, and particularly tangible non-screen-and-keyboard based technology, can play a
unique role in supporting emergent writing literacy activities. In the remainder of this article, I propose a
model of technology for literacy, whereby four essential traits allow technology to effectively scaffold
literacy. Those four traits are embodied in a suite of technologies that I have developed, called Story
Listening Systems, which I then describe. The four traits are to (a) depend on children’s oral storytelling
skills to bootstrap literacy, (b) introduce peers as playmates in the system or with the system, (c) invite
the kind of embodied play away from the desktop that is most comfortable for young children, (d) allow
children to construct their own personally meaningful content. In order to evaluate the effectiveness of
these systems, I rely on three crucial predictors of literacy: (i) using decontextualized language (language
removed from its original context, and reworked for a new audience), (ii) gaining metalinguistic
awareness, (iii) collaborating with peers to make meaning. In addition to having an effect on these three
Story Listening Systems - 3 -
predictors, Story Listening Systems can also, I believe, impact children’s learning of literacy skills in a
fourth way, and that is by participating in a broader definition of emergent literacy in such a way as to
successfully scaffold literacy for children whose pre-literacy behaviors don’t fit the mainstream model.
2. Background
Wells (1981) describes three broad phases of language development. The first involves the discovery that
language is a pattern of sounds that takes on meaning and purpose. In the second phase children come to
understand the social aspects of language, that assumptions and values are encoded in particular
linguistic representations and that these values are specific to a particular community (e.g., the style of
language used on the playground is different from the style of language used at school). These first two
phases involve a close link between speaking and doing. Wells describes a third stage of development
that involves the creation and manipulation of language designed for an audience that is spatially or
temporally separated from the author. Wells calls this stage “literacy,” and defines it as the ability to
communicate with an audience not in the immediate physical or temporal context.
2.1. Decontextualized Language
Literacy, then, is the ability to make meaning for others across space and time. Only in this stage does
language have permanence. It can be use for reflection, memory and sharing meaning with others not
currently present. Children’s use of literate language occurs, in a sense, the first time meaning is
separated from context. It is in this way that language becomes “decontextualized,” or removed from its
original context, and reworked for a new audience. Literacy as “decontextualized” language use is a
central aspect of the literacy definition used in the current proposal.
Another way of thinking about the aspects of literacy that will concern us here has to do with the
distinction between “inside-out literacy skills” and “outside-in” literacy skills (Whitehurst and Lonigan,
1998). The former consists of skills on which are based the actual act of writing: such as phonological
awareness, knowledge of how to form letters, and the kinds of punctuation involved in writing. The latter
consists of skills which relate to the function and features of writing in the world, and it is those skills
that will concern us here. Children’s readiness for the outside-in aspects of writing literacy begins in
play and storytelling activities that don’t explicitly involve the decoding or creation of text. One of the
most important outside-in writing literacy skills is the ability to deal with decontextualized language
(Snow, 1983). Decontextualization includes the ability to maintain cohesion and reference (Gee, 1985;
Michaels, 1986; Peterson, Jesso et al., 1999) through manipulating linguistic devices such as (1) tense
and temporal adverbs (e.g., when, after), (2) connectives (but, and, so), and (3) referring expressions
(she, one) (Nelson, 1996). In fact, these markers are precisely what teachers look for in school-based
discourse (Michaels, 1986). Storytelling is a perfect place for children to practice talk about the “thenand-there” (Scarlett and Wolf, 1979) as they learn to distance objects, actions, and feelings both in
thought and in language (Nicolopoulou, 1996). And structuring experience as stories (“narrative
thought”), and telling those stories are skills that come naturally to children, who have “an abundant and
early armament of narrative tools” (Bruner, 1990: 79).
This description counters the view that the fundamental differences between speaking and writing mean
that their acquisition is quite separate. Sulzby (1996) shows that children do not acquire oral and written
language in a linear sequence but instead that the two are intertwined. Her model of emergent literacy
describes children’s language acquisition as the result of co-occurring competency with both oral and
written media and the contexts in which they are used. For young children, then, encouragement to both
produce oral language for an audience that requires sophisticated use of decontextualized linguistic
devices, and to begin to write, may be the most successful approach to integrating the insights of
emergent literacy into the classroom.
Story Listening Systems - 4 -
2.2. The Role of Peers
The “somebody” for whom language is produced is often peers. Although parents certainly introduce the
importance and possibilities of writing to children (through writing thank-you notes to relatives, grocery
lists, and so forth (Hall, 2000)), it is through interaction with peers that the conception of the other as
audience, with its attendant demands for decentered and decontextualized language, matures. In emergent
literacy both the cognitive and the social roles played by peers are vital (Damon and Phelps, 1989). The
inter-dependence between peers is unique in how it pushes children to take the perspective of the other
and to function at a higher level than otherwise. In fact, when two peers collaborate, the simple
juxtaposition of their actions appears to allow the peers to modify their understanding of their own
actions, through appropriating the perspective of the other peer (Rogoff, 1991). That is, to apply
Rogoff’s notion to emergent literacy, the very fact of telling a piece of a story that follows after the piece
told by one’s peer, allows both peers to gain a new understanding of the meaning of their words in the
context of the story (Doise, 1990: 46). But peers also serve quite explicit instructional roles with one
another in literacy contexts. Neuman and Roskos (1992) observed children engaged in instructional
conversation with their peer, negotiating, and coaching each other’s literacy activities. Unlike in adultchild instructional conversations, children instructing one another often reversed roles and attributed the
role of the more capable peer according to the purpose of the play at hand. Similarly, Stone and Christie
(1996) observed children helping each other by modeling, assisting, directing, tutoring, negotiating,
affirming, and contradicting each other in literacy activities. In fact, these functions serve a concrete role
for both parties: Engaging in peer talk that served a wide range of social functions (from asking questions
and initiating text sequences to playing) was positively correlated with the change toward writing in the
third person (Daiute, Campbell et al., 1993).
2.3. Metalanguage
Finally, essential for writing literacy is awareness not just of language but also of metalanguage.
Metalinguistic awareness is the ability to attend to and reflect on the nature, structure and function of
language. Even very young children who are just learning to speak demonstrate a kind of metalinguistic
awareness when they play with the sounds of language. By the time they are two years old, children
begin to play with the words that make up language, often in “conversations” with themselves, when they
are alone (Nelson, 1989). Slightly older children engage in similar behavior, but in collaboration with
others. Children of this age derive tremendous pleasure from rhyming words (“you silly,” “no, you
pilly”) or words that sound similar (adult: “Indians lived in a teepee”; child: “pee-pee!”). Slightly more
complex kinds of metalanguage that imply actual reflection on language, such as the use of puns, which
demonstrate a metalinguistic awareness of the multiple meanings of words, have been shown to be
important for reading and for written language development (Hiebert and Raphael, 1998). Around the
age of four, children begin to acquire metalinguistic awareness of pragmatics – knowledge of different
communicative contexts, and the ability to describe language (Shatz and Gelman, 1973; Gombert, 1992).
Children of this age begin to mimic accents and different speech styles. In fact, traditional ‘reading
readiness’ measures predict reading ability less well than children’s use of more explicit metalinguistic
terms such as say, write¸and talk during their spontaneous pre-school play (Pellegrini and Galda, 1993).
Three- and four-year old children’s use of narrative perspective in “reading” a storybook has been
correlated with later academic competence in math (O'Neill and Pearce, in preparation). These terms, and
others that introduce reported speech, or bracket instances of embedded language of one kind or another,
are important in part because they demonstrate a growing ability to conceive of language as something
that is produced by somebody for somebody.
Story Listening Systems - 5 -
2.4. Broadening the Definition of Emergent Literacy
However, the particular features of narrative most often prized by teachers (e.g., explicit use of linguistic
markers and logical sequences) are encouraged primarily in mainstream culture (Gee, 1985; Heath, 1986;
Michaels, 1986). For example, traditional African-American oral narrative does not depend on the
restricted temporal and causal chain ordering conventions of “school-literate” narrative (Michaels, 1986).
And yet genres of language play and of storytelling exist in all cultural contexts. Despite descriptions of
the multicultural aspects of storytelling activities, and some descriptions of their link to children’s
literacy and other aspects of development (Labov, 1972; Lee, 1992; Miller and Hoogstra, 1992), virtually
no attempt has been made to integrate their benefits into the classroom (see (Pinkard, in press), described
below, for a notable exception). Indeed, the specific kinds of language play demonstrated by AfricanAmerican children is sometimes devalued and belittled to such an extent that African-American children
lose their desire to participate in the classroom (Michaels, 1981). This is the more the pity since, as
Heath (1986) explains, an exposure to many different speech genres is essential for all children. Heath
points out that none is inherently more elementary than another, and all students could benefit from
having exposure to many types of language activities and genres. And when teachers respect the
different discourse and literacy practices in the home, children can be highly successful in school-defined
literacy acquisition (Katsarou, 1992).
Many of the activities that we have described as contexts in which literacy emerges are play among
children. Play among children of this age group involves the whole body, and yet when literacysupporting technology is found in the classroom, it typically resides in a desktop computer where an
individual child is placed in front of a computer monitor and a keyboard. This makes it difficult both for
children to collaborate, and for them to involve their bodies in play. We believe non-screen-andkeyboard technology is important in scaffolding collaboration, and in encouraging verbal, non-verbal and
embodied aspects of children’s language play. Research in educational technology has not, however,
tended to look at the speaking-writing continuum, nor the benefits of storytelling. In the next section we
describe research that has come close to the goals laid out in the current proposal.
2.5. Relevant Previous Technological Research
Some other research has addressed technology for literacy, although more previous work has
concentrated on technology that supports children’s reading rather than their writing. For example,
Project LISTEN (Mostow, Aist et al., in press) has demonstrated impressive results with an automated
(non-embodied) Reading Tutor that displays stories on a computer screen, and listens to children read
aloud. Mostow found that over a period of 4 months, children using the Reading Tutor gained
significantly more in passage reading comprehension than children using commercial reading software,
or engaging in classroom reading activities. A number of companies (for example, LeapFrog and
Scholastic) have also targeted reading literacy, although for the most part their products have addressed
the “inside-out” skills of phonetic and graphemic awareness.
As described above, oral storytelling takes place in households of many different cultural backgrounds,
but teachers are often unable to use emergent literacy activities that differ from their own to scaffold
children’s classroom skills. Pinkard introduced the Lyric Reader software (Pinkard, in press) built upon
the experiences and background of low-income African-American children. Two Lyric Reader
applications, Rappin’ Reader and Say Say Oh Playmate provide “contextualized reading instruction”
with reading material drawn from children’s prior knowledge of song lyrics. After children used these
technologies for two sessions, Pinkard found that while European-American children’s pre- and post-test
raw scores in vocabulary tests were greater than African-American children, at all grade levels, AfricanAmerican children demonstrated a greater percentage word gain than European-American children. Like
Story Listening Systems - 6 -
the Reading Tutor, Pinkard’s work depends on a traditional screen-and-keyboard computer that resides in
a classroom.
A number of technologies to support children’s storytelling have been developed. For example, Druin
and her colleagues have designed a number of interfaces for and with children focused on supporting
children’s storytelling play (Druin, Stewart et al., 1997; Benford, Bederson et al., 2000). Druin’s work is
strong in its attention to collaboration among children and argues that by including children in the design
process, technologists can create storytelling applications that are more representative of the authorship
issues children find salient. Hayes-Roth’s Improvisational Puppets System (Hayes-Roth and Gent, 1997)
provided an environment where children can act out plays. MOOSE Crossing (Bruckman, 1995)
encouraged children to construct a virtual environment, using narrative descriptions, in which they could
interact with one another. Graphic StoryWriter (Steiner and Moher, 1992) let children manipulate
multimedia objects as a way of supporting storytelling. Another multimedia authoring tool encouraged
children to treat videos, photos, drawings, texts, sounds and cartoons as “electronic building blocks” that
represent segments of time in a story (Kim, 1995). Druin’s newer work has begun to embed storytelling
technologies into everyday objects; otherwise, her older systems and the other systems described here,
reside in desktop computers. And, in none of these systems is the link between storytelling and literacy
addressed.
The Cognition and Technology group at Vanderbilt University, has targeted reading, writing, and
storytelling in an innovative literacy series that has had considerable success in the classroom. Children
in primary grades first listen to and watch an “anchor” story presented in a video format, fostering oral
comprehension skills. The stories end with a challenge to the children that makes it important for the
children to write a book about the story they have just seen and heard. They then use specially designed
“book making” software to produce a reconstruction of the story that they can print out in book form and
“read” to their peers and family. Comprehension is supported by having the students use storyboarding
technology to decide on the temporal and causal sequence of events in the book. The story reconstruction
is done orally by having children record what they want to say as well as in written form by typing in the
words for each page. Children can also create their own stories in the book making software. Note taker
technology helps them write stories or nonfiction reports from scratch. The project has demonstrated
gains on reading comprehension and in various measures of written story production (e.g., word fluency,
sentence fluency and story complexity) among high-risk low-SES first-grade children (Cognition and
Technology Group at Vanderbilt, 1998).
Some Intelligent Tutoring Systems (ITS) have examined the role of peers or partners in learning. For
example, Chan and Baskin (1988) proposed the notion of a “learning companion system.” The learning
companion was an artificial student who interacted with the real student while both learned under the
guidance of an intelligent tutoring system. The learning companion was designed to be at about the same
level as the real (college) student, and both the student and the companion exchanged ideas while being
taught by the computer teacher. The relationship between the two students was designed to include
cooperation, competition, modeling, co-teaching, and observing. By having the two tasks – learning by
being tutored and tutoring, learning companion systems offer a learning protocol that is similar to
“reciprocal teaching” (Palincsar and Brown, 1984) where children take both the teacher’s and learner’s
role. While their preliminary results did not show significant improvements on problem solving tests,
Chan and Baskin’s interviews revealed that the students enjoyed teaching an agent over a real student
because they felt it was like a game. It is possible that a more complete model of peer relations would
maximize the effects of this kind of ITS. In the Teachable Agent project (Brophy, Biswas et al., 1999),
children learned ecology by teaching an (non-embodied) agent about the subject. Brophy et al. found
that children who studied in order to teach the agent did better on the post-test than control children who
studied just for the test, as the students who prepared to teach spent time trying to understand “the why”
of what they were learning. Animated Pedagogical Agents (Johnson, Rickel et al., 2000) are intelligent
Story Listening Systems - 7 -
tutoring systems that have been given an embodied representation, which allows them to demonstrate
visually how to perform tasks, and communicate using language, hand gestures, and facial expressions.
Empirical evaluations of these systems have yielded mixed results. For the most part, subjects (the vast
majority have been college students) report enjoying the learning experience more, but comprehension
and recall have not been improved by the presence of a visual interface, or image of the agent (Dehn and
Mulken, 1999; Moreno, Mayer et al., 2000). In these systems, however, it is unclear whether the
contribution of embodiment has yet been capitalized upon. That is, although these pedagogical agents can
demonstrate behaviors by pointing and demonstrate enthusiasm by facial expressions, they do not yet
distribute functionality to language and to embodied behaviors in maximally human-like ways
Although certainly not a new technology, a particularly intriguing result from a study looking at
contextual effects on the production of narrative language demonstrated that stories children told over the
phone were longer, more detailed, and more vivid than those told face-to-face (Cameron and Wang,
1999). Cameron and Wang suggest that the mediation of the telephone encourages children to not
decontextualize, but – more intentionally – recontextualize their language for an audience. Note that, as
Naigle reports (Singer, 2001), and despite popular claims to the contrary, television was never shown to
have a positive effect on language learning (meaning, the development of grammar, and not just
vocabulary). Television viewing is more active than many people might suppose (Bickham, Wright et
al., 2001), but televisions simply do not reply to children. Unlike television, however, new technologies
can encourage active construction of language, and their mediating properties, I believe, can be used to
encourage metalinguistic reflection. It is for this reason that I believe that Story Listening Technologies
can encourage children to make gains in language skills. I believe that story listening systems will
encourage children to recontextualize their language for an interlocutor, in a similar way to that
demonstrated by Cameron and Wang – whether that interlocutor be co-present or temporally/spatially
distal, real or imagined.
In sum, narrative is of paramount importance to literacy. In order to support emergent literacy, children
need to author their own personally and culturally meaningful content in a way that is representative of
the oral-written emergent literacy continuum, but that focuses them on key aspects of decontextualized
language use. These goals can be achieved through storytelling and story writing, activities that are also
important in children’s broader cognitive, communicative, and linguistic development. Many schools are
beginning to recognize the creation and manipulation of oral and written text as key aspects of literacy
education. In parallel, teachers are being encouraged to incorporate technology into classroom activities.
Technology, and particularly tangible non-screen-and-keyboard based technology that can support
children’s language play involving the whole body, may play a unique role in supporting emergent
writing literacy and otherwise making a bridge between oral and written language, both in the classroom
and outside.
3. A Model for Technology and Literacy Development: Story Listening Systems
With the exception of the work of Goldman and colleagues, previous work on literacy technologies has
for the most part eschewed writing in favor of reading. And, even within reading, most of these
technologies approach literacy from the inside out – graphemes, phonics, and punctuation – rather than
the more social and functional approach. Technologies for literacy and for storytelling have been
approached separately, and thus the personal and epistemological connections that can arise from
constructing one’s own language have been ignored (Resnick, Bruckman et al., 1996), as has the
possibility of using technology to bootstrap academic competence off of a skill that the child is already
good at. Finally, technology to support the collaborative social and peer context of emergent literacy,
such as pretend play or oral storytelling in pre- or non-school contexts, has not been explored.
Technology of this sort would have to move away from the desktop screen-keyboard interface, since
young children primarily engage in collaborative oral storytelling and fantasy play using their whole
Story Listening Systems - 8 -
bodies, and may be utterly slowed down by the constraints of a keyboard and mouse. Overall, then, with
rare exceptions, it appears as if technological innovation has not kept up with innovation in literacy
learning theory, and that the four key features of the literacy development model named above and
underlined as important by researchers in (non-technological) approaches, have not found their way into
technologies for literacy development.
What is needed to support the outside-in components of literacy is a social context – a feeling of being
understood, and having an authentic environment in which to pursue one’s communicative goals. Of
course, technology will never replace a genuine social context – peers and adults who care. I seek to
improve on current technology for literacy by extracting those aspects of the social context of peer
interaction and of adult interaction that facilitate literacy learning, and use enabling technologies to
perform them. Thus, the SLS model relies on two features of childhood and two features of new
technology: children's inherent ability and desire to tell stories which may contain within them the skills
required for writing; the fact that the support children provide to one another in the narrative play arena
may be as good as, or even superior to that provided by adults; new technology's ability to be removed
from the old screen-and-desktop context; and its capacity to support children as producers as well as
consumers of content. I believe that, by proceeding along these four axes, based on essential properties
of childhood and possibilities of new technology, we will create an effective context for the acquisition
of literacy skills. Story Listening Systems share these four essential traits. (a) They evoke storytelling
from children, and use children’s oral storytelling skills to bootstrap literacy, (b) they encourage peer
play, and sometimes even embody peers as playmates in the system, (c) they do not require children to
learn desktop screen-keyboard-and-mouse interfaces or to sit at a desk before they can engage in normal
children’s behavior; instead they are built into children’s toys, and thus invite the kind of embodied play
that young children engage in, (d) rather than constraining the kind of stories that children can tell or
listen to, they ask children to construct their own personally meaningful content. I predict that systems
possessing these four characteristics can allow children to succeed at three crucial predictors of literacy:
(i) using decontextualized language, (ii) gaining metalinguistic awareness, (iii) collaborating with peers
to make meaning. They can also, I predict, succeed at (iv) broadening the definition of emergent literacy
in such a way as to successfully scaffold literacy for children whose pre-literacy behaviors don’t fit the
mainstream model.
One of the benefits of the approach is that the SLS allow abundant access to the child’s process of
storytelling – they allow, therefore, for study of the learning mechanism that underlies the SLS. Each
stage in the child’s interaction with the object is saved for analysis. As a result, comparison of children’s
uses of different SLS, and of SLS with other kinds of literacy instruction can be compared to evaluate
each of the features that makes up the model, and thus evaluate the design features for literacy
technologies.
4. Six Story Listening Systems
In what follows, I first describe three early “story listening systems” (Renga, Rosebud, SAGE) designed
to look at the general role of technology in children’s storytelling, and that we have evaluated through
informal, ethnographic, and quasi-experimental means. Next, I turn to three story listening systems
(TellTale, StoryMat, Sam) that have explicitly addressed the value of “toys that listen” for fostering
children’s emergent literacy skills. These latter three systems target those aspects of language interaction
that are a source of literacy-relevant and literacy-necessary skills. Finally, I will describe future work that
supports children in an explicit transition from oral language to writing.
Story Listening Systems - 9 -
4.1. Renga
In Renga (Cassell, 1995) we examined the collaborative
and community-building aspect of storytelling. Our
objective was simply to experiment with the technical
capabilities of the web as support for the social activity of
storytelling. On the basis of the kind of round-robin
storytelling that children often engage in in first grade
classrooms, Renga (from the Japanese word meaning,
“linked poem” or “linked image”) encouraged children to
add a sentence to an ongoing story on the web, and in
doing so to become a part of a storytelling community. In
our initial introduction of Renga, we invited the
participation of every school in the world that had web
presence (214 schools, in October, 1995!). Children from fourteen schools in eleven countries accepted
our invitation and used Renga to collaboratively tell stories during one 24-hour period in October 1995.
Our experiences with Renga during that 24-hour period, and children’s subsequent reactions during later
informal use of the system, led us to realize that the nature of storytelling collaboration among children
requires a social as well as task-oriented component. Several children explicitly asked us how to find out
who else was “there.” Our interpretation of this reaction was that it is not enough to collaborate on a
task; the collaborators’ social presence (Short, Williams et al., 1976) must somehow be felt. To this end,
we added a function so that when children clicked on any sentence in the story, they were able to see
information about the author -- name, country of origin, hobbies – and at any time the names of the 5
most recent collaborators were listed in the storytelling “room.” Renga continued to be used over a year
or two period, however, after that point scalability became an issue due to the increasing numbers of
children on the web.
Renga succeeded in supporting a kind of collaborative storytelling among children across distance, and
succeeded in focusing children on the task of constructing local linguistic links between the sentences
that they wrote, and the previous sentence on the page. But the nature of the collaboration was not
inspired by any particular topic – the story could be about anything – nor did it have any overarching
structure and so it typically devolved into rather unrelated sentences 100 or so lines down. In addition,
there was no facility for rewriting, editing, reflecting on one’s own language or the language of one’s
peers, nor was there any audience – notion of who the story was written for. To address these
limitations, we designed Rosebud.
4.2. Rosebud
The Rosebud system (Glos and Cassell, 1997) was designed
to support collaboration between multiple children, and
between a child and his/her stuffed animal; also to encourage
writing, and reflecting on writing. In Rosebud the computer,
in the form of a screen-based storybook, recognizes
children's stuffed animals (via an infrared transmitter in the
toy, and receiver in the computer) and invites the child to
type a story about the stuffed animal into a magical
storybook. With each play session, the child accumulates
more stories about the stuffed animals in the personalized
storybook. Once a story has been written, the child is
encouraged to “tell more,” to revise, and to expand.
Rosebud supports storytelling not only by one child and one stuffed animal, but also by multiple children
each with his/her own stuffed animal, working together. Rosebud allows collaboration by allowing
Story Listening Systems - 10 -
multiple-toy use and multiple-author storybooks, so that several children could write a story together
about all of their stuffed animals. We instrumented a menagerie of stuffed animals, including highly
“mediatized” figures such as Winnie the Pooh, and anonymous figures such as an elephant. Use of the
system by 7 to 9 year old children demonstrated that, regardless of the identity of the stuffed animal,
highly personal stories were likely to be told, as we can see in the following story told by one 9 year old
girl playing with a Pooh Bear and Eeyore animal:
Once upon a time a very long time ago there were two friends named pooh
bear and eeyore. They liked each other very much and let nothing get in
there way. One day eeyore felt sad because his mother passed away. So
he went to his good friend pooh bear to ask for advice. Eeyore asked
pooh bear and pooh bear said to take it easy and relax. Eeyore said ok
I will.
Children’s most frequent request was to allow them to record their story into the stuffed animal so they
could take the animal with them, and share their stories with their friends. We implemented this
function, but their use of the system made clear to us the importance of moving the interaction away from
the desktop computer. In addition, despite the edit and revision features of the software, children were
not particularly interested in rewriting – there seemed to be no reason intrinsic to the system for them to
reflect on what they had written, and to rework it. To address these limitations, we designed the SAGE
system, where the structures of discourse were explicit in such a way as to encourage children to be
reflective about language use.
4.3. SAGE
The goal of SAGE (Bers and Cassell, 1998) was to invite children to design their own ideal listeners, and
in doing so to reflect on the speaker and audience aspect of storytelling. The SAGE (Storyteller Agent
Generation Environment) story listening system encouraged children to tell personal stories by giving
them a library of computer-based wise old sage storytellers to choose from, each of whom listened and
then responded with a relevant tale of his/her own. Children could choose to speak to a rabbi, who
responded to children’s stories by saying “what, you think you’re maybe the first person this happened
to? Let me tell you a tale” and then chose a relevant story from a database of Hasidic tales; a Buddhist
scholar who always had a relevant Tao proverb; or a French grandmother who replied with one of the
fables from La Fontaine. Older children (we tested the system with ten and eleven years old) could also
create their own sage storytellers and to design storytelling interactions for their peers.
In order to support children as constructors as well as
consumers of the storytellers, we implemented a visual
programming language that allowed children to design and
program: (1) the scripts used by the storyteller to respond to
users, (2) the conversational structure or flow of the
interaction, (3) the database of stories offered in response by
the storyteller. We saw children, over the course of a
weekend workshop become able to observe and repair
breakdowns in conversational interaction. For example, one
child built a “wise old person” based on Shaquille O’Neal,
the basketball player. In a demonstration of the system,
when Shaq asked the child’s father, “I guess you are around
11, am I right?” and the response was, “No, I am 45,” Shaq crashed. The child quickly realized that he
hadn’t programmed a branching node to allow the possibility of a “no.” He went back to the authoring
mode and added the branching to the conversational structure. This process appeared to engage children
in the exploration of notions of communicative decentering. As one eleven year old told us in the postworkshop interview, “First I learned how to put myself in another place and pretend I am someone else,
Story Listening Systems - 11 -
and I learned how hard it is [. . .] But really I talked to myself and I learned more about myself [. . .] My
stories are sad, but if you wante to hear a funny story you can play with Waloompa the alien. When you
have a problem it is sad and that is why my stories are sad." The SAGE system has been given a web
interface so that children may design story listeners for children around the world.
SAGE was quite successful in encouraging metalinguistic awareness, and also reflection on the activity
of storytelling – its structure and function. In order to be engaged in this reflection, on language and on
storytelling, however, children needed to master a programming language that was more difficult than we
expected. Even the eleven year old children we worked with had difficulty with the task, and children of
this age group are less in need of scaffolding for metalinguistic awareness than are much younger
children. Our experiences with the SAGE system led us to turn our attention to much younger children,
systems that are simpler to use, and more explicit scaffolding of pre-literacy behaviors.
4.4. TellTale
TellTale (Ananny and Cassell, 2001) was the first
SLS designed explicitly to engage children in the kind
of discourse that is needed for writing literacy.
TellTale illustrates by its form an important concept
of writing: units of discourse must hang together
somehow, and then be connected to other units, and
there must be a beginning and an end. In this sense TellTale resembles the manipulatives (Resnick,
Martin et al., 1998) used to support children’s exploration of math and physics. TellTale is a caterpillarlike toy with five modular body pieces and a head. Children can press a button on each of the five body
pieces to record 20 seconds of audio. The child can then press a button on that body piece to play back
the audio. The body pieces detach from one another and children can arrange and rearrange them in any
order. At any point the child can attach the toy’s head to the body to hear the entire audio story played in
sequence. Any body piece can be re-recorded, or re-arranged.
4.3.1. Metalanguage. Initial informal observations of children playing with TellTale revealed
that the artifact appeared to affect how children reflected on their story language and how they
experimented with story structure. Four stories recorded one after another by one 6-year old illustrate
this pattern and are shown in the table below. The first story this 6-year-old girl recorded is a complete
and coherent narrative. She only used two of the five body pieces, saying she “didn’t need the other
ones.” In this first story, then, the form of the caterpillar played no role in the form of her story. In her
next story she used all five body pieces by putting shorter story segments in each, which involved at least
some advanced planning about the structure of the story. The third and fourth stories were recorded in
response to a question the experimenter posed about why she liked or disliked TellTale. She responded
that she liked “TellTale a lot because you can split things up in different ways – see?” and then
proceeded with the third story – in which she predominately recorded a single word into each body piece
– and the fourth story – in which she recorded a single clause into each body piece. Note that the fourth
story is based on the third but conveys more information. The third and fourth story demonstrate this
child’s ability to reflect on the structure or segmentation of the story, and how to map it to the structure
of TellTale. Thus this child is engaging in metalinguistic awareness of discourse segmentation.
Story Listening Systems - 12 -
story
Body
Piece
1
th
4 story
rd
3 story
2
nd
story
st
1
recorded
2
3
4
5
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
Audio recorded into body piece
Once upon a time there was a little boy who was sleeping. The next morning he lost his pet frog. Then he
went looking and … [with rising intonation]
He looked in a hole and the dog looked in the beehive. He looked I think under a rock and an owl flew
over. And then he was calling his frog’s name. He fell into the river. He was looking and he said “shhh”.
Then he found his frog. The end.
Did not use
Did not use
Did not use
Once upon a time there was a unicorn. And he was so pretty.
And when he touched its horn a jewel came. And he touched anything and a jewel came
And then it lay and it wanted me to get on its back.
Then I rode to its castle.
And saw a real live princess. The end.
Once
Upon
A
Time
There was a unicorn.
Once upon a time
There was a unicorn
Named Crystal
And she liked to play hide and go seek with people. And she was a good person.
The end
These first informal observations led us to design a study to look more explicitly at the development of
an awareness of narrative segmentation and structural language, as embodied in TellTale’s segmented
form.
4.4.2. Decontextualized Language. We constructed a control condition TellTale where only one
piece recorded audio, and it contained 100 seconds of audio (the same amount as the entire original
TellTale). 14 children playing alone with the unified unit (UTT) or segmented unit (STT) TellTales
were invited to record stories (in a room with no adult present). We examined the use of connectives and
other kinds of macrostructure markers at the boundaries between body parts. Connective use by children
can be divided into local chaining and macrostructure markers (Peterson and McCabe, 1991).
Segmented TellTale
Unified TellTale
(Child aged 6 years, 11 months)
(Child aged 6 years, 7 months)
BP#1: “Once upon a time there was a unicorn. And he
was so pretty.”
BP#2: “And when he touched its horn a jewel came. And
he touched anything and a jewel came.”
BP#3: “And then it lay and it wanted me to get on its
back.”
BP#4: “Then I rode it to its castle.”
BP#5: “And saw a real live princess. The end.”
“The caterpillar had just got home. He didn’t
know where he was. He asked the horse where
his mother was. The horse said …said he was …
she was … at … the … leaves she … it … she
said …”
Stories told with STT were longer than those told with UTT: STT stories were an average of 72 words
per story and 40.5 seconds per story whereas UTT stories were an average of 42.1 words per story and
34.2 seconds per story (N was too small in this study to report inferential statistics). Stories told with
Segmented TellTale (STT) had fewer false starts than those told with Unified TellTale (UTT) indicating
Story Listening Systems - 13 -
that the segmented body pieces may allow children to plan their utterances off-line. Stories told with
STT also had contained more conjunctive phrases (and, then, however, when, while, after, later, so,
therefore, one day) per word (.1 conjunctions/word) than those told with UTT (.06 conjunctions/word).
And when conjunctive phrases did occur in STT, they tended to occur at body piece boundaries,
indicating that children treated body pieces as story units, linking them with connectives.
In both UTT and STT conditions children tended to tell stories with classic beginnings (e.g., “once upon
a time”) but only in the STT condition did children also consistently finish their stories with classic
endings (e.g., “the end”). Stories told with UTT tended to end in either false starts or long pauses
indicating that children may have been having difficulty planning the next utterance.
TellTale’s segmented interface, then, seems to help children tell stories that are longer, more cohesive
(containing fewer disfluencies and more conjunctions) and with more traditional beginnings and ends.
The skills children practiced while playing with the segmented version of TellTale (planning, chunking,
revising) are very similar to those that are required for written literacy. These findings suggest that, with
respect to segmentation, TellTale encourages children to tell oral stories in ways that are similar to how
they will eventually construct written texts.
4.4.3. Multiple Discourse types. TellTale does not restrict in any way (other than length) the
kind of discourse that can be recorded. We carried out an evaluation in two schools in Dublin, Ireland to
examine how TellTale may support the language play of children from different socio-economic strata
(SES). Specifically, this evaluation investigated whether children of high- versus low-SES establish
cohesion within oral narratives using different strategies. A total of 22 children participated: 5 low-SES
dyads (10 children) and 6 high-SES dyads (12 children). All children were invited to “tell stories with
this storytelling toy.”
Despite the explicit instructions, in general we found that children in the low-SES condition used
TellTale in unanticipated ways: they were less likely to tell complete stories, and more likely to record
song fragments, dialogues and rhyming sequences into the body pieces, or record an amusing quotation
into each body piece. Children from both high- and low-SES groups consistently recorded utterances
that contained story events (in both conditions, approximately 75% of all children’s recordings contained
at least one event). Children from both high- and low-SES groups also used connectives at TellTale body
piece boundaries but children from high-SES consistently used more connectives. In terms of the process
of telling stories, interestingly, both high- and low-SES children incorporated both narrative (characters,
places, actions) and syntactic (exact word phrases) aspects of their partner’s utterance into their own.
But low-SES children tended to make incorporations simultaneously (i.e., during co-occurring
recordings) whereas high-SES children tended to incorporate each other’s content across consecutive
recordings. Children from both high- and low-SES engaged in cohesion with the previous child’s
segment through paralinguistic means (e.g., rising and falling intonation) and non-verbal means (e.g.,
gestures and eye-gaze). But children from the low-SES group were less likely to use cohesive devices at
all, or they used only paralinguistic and non-verbal strategies, rather than explicit connective chaining
with the previous child’s segment.
Overall, children from different socio-economic strata tended to engage in slightly different behaviors
during collaborative storytelling. Specifically, an initial analysis may interpret low-SES children’s high
percentage of co-occurring utterances and low percentage of syntactic connectives as an indication that
they are less able to engage in good turn-taking behavior and be aware of their co-participant. However,
this may not be the case for two reasons: children from the low-SES group appear to be using nonsyntactic, paralinguistic and non-verbal strategies to indicate cohesive chaining during story construction.
Also, despite the high percentage of co-occurring utterances in low-SES children’s recordings, these
children consistently incorporated elements of their partner’s utterances simultaneously. Children from
high-SES tended to establish coherence using syntactic connectives between consecutive recordings.
Story Listening Systems - 14 -
4.5. StoryMat
StoryMat (Cassell and Ryokai, 2001) was designed to encourage exactly the aspects of 4 to 6 year old’s
collaborative reciprocal storytelling that have been shown to be related to later literacy (Dickinson and
Tabors, 1991). In StoryMat the voices of previous children who have played with the toy serve as peer
collaborators, egging children on to continue producing narrative language, and providing a kind of
virtual audience for the children’s words. We hypothesized that even so thin a representation of a peer as
a disembodied voice would, as described by Dyson (1993) imbue the child’s symbolic acts with social
meaning in such a way as to drive the acquisition of those acts of symbolicization.
StoryMat was entirely removed from the desktop, and provided
children with (virtual) peer collaborators. The implementation of
StoryMat was a soft cloth quilt with appliquéd figures on it – a
familiar toy for young children, one that evokes narrative and
whole-body play. Squeezing one of the small stuffed animals
provided with the quilt triggered the system to start recording the
child’s narrating voice and the two-dimensional coordinates of
the stuffed animal. When the child let go of the stuffed animal,
the coordinates and the voice were combined into a movie file
and saved in the computer to be played at the appropriate
location on the mat. When new input was subsequently
encountered at the same place on the mat, the movie file was then automatically triggered and played
back via a projector mounted above the mat, and heard through a pair of speakers next to it. The current
child could then tell her next story. Sometimes she might come up with a continuation of the story she
just heard. Or she might continue telling her own story, incorporating some story elements from the story
she just heard. In this sense, StoryMat is a kind of imaginary playmate (Taylor, 1999), but who also
mediates collaborative storytelling between a child and her peer group.
Our evaluation of the StoryMat system concentrated on three kinds of emergent literacy activities among
early school-age children. First symbolic transformations of real objects into fantasy story objects,
important because of the link between symbolic transformations and later writing literacy (Pellegrini &
Galda, 1993). Second, incorporations of story elements from the story of a collaborating peer, important
because of the increased complexity found when children incorporate aspects of each other’s stories
(Corsaro, 1992), and the role played by peers in encouraging more sophisticated linguistic devices
(Bokus, 1992; Preece, 1992). Third, the use of different narrative voices, because of the essential role of
perspective taking in the development from oral to written literacy (Peterson and McCabe, 1983), and the
specific role of narrative voice in acquiring more sophisticated kinds of story structures (Auwarter,
1986). These decontextualized language measures were taken from previous work cited above, and were
therefore well validated. 36 children between the age of 5 and 8 participated in the evaluation study.
Children were randomly assigned into one of two groups: 1) a StoryMat group, who played on StoryMat
and 2) a control group, who played on an identical mat, without the responsiveness (the “passive mat”).
In each group, 6 subjects played alone and 12 subjects played with another playmate, resulting in 6 dyads
and 6 singles in each group. In all conditions, the experimenter’s instructions were fundamentally the
same: “let’s pretend that we are living in this world. Will you tell me stories that happen in this world?
First I’ll play with you some. But I’m going to leave the room in a little while so that you can be alone to
tell your stories on the mat.” Children on the StoryMat condition were given one additional instruction:
“Do you see a button here? (Show the button on the stuffed rabbit) You have to hold down the button
while you tell your story so that the mat knows you are telling your story. And you let the button go
when you are done so that the mat knows you are done with your story.” The experimenter always told
the first story, and in the StoryMat condition the experimenter then sat and listened to the story that the
mat gave back in reply. In all conditions the child was then invited to tell a story, and then the
Story Listening Systems - 15 -
experimenter left the room. Thus, in all conditions children heard an equal number of practice stories
before the actual session began. No additional instructions were given as to the functioning of the
system, nor what was expected of the child. The experimenter was absent for the entire play session, as
many studies have shown that children interact less with one another when they are in the presence of
adult observers.
Results demonstrated that StoryMat encouraged more symbolic transformations in the children’s stories
than the control condition for both children who played on the mat alone and with a co-present peer (F
(3,20)=9.7, p<. 01). Children playing on StoryMat also incorporated story elements from the stories
offered by StoryMat in the similar way as they did from real life peers (F (3,20)=3.49, p < .05). Finally,
children playing alone on StoryMat more often took the more narratively advanced role of narrator role
(72%) than of character (28%), while the control group of children playing alone acted in character role
(95%) much more frequently than in narrator role (5%). Some of the advantages that accrue to children
playing with peers, then, seem to accrue to those children playing with the StoryMat system.
4.6. Sam the Castlemate
The Story Listening Systems described above incorporate some implicit aspects of peer support for
narrative performance. However they contain no embodied representation of the other (virtual) children
with whom the child is playing. Embodied Conversational Agents (ECAs) can serve this role. ECAs are
life-size human-like graphical computer characters projected on a screen that are able to engage in faceto-face dialogue with a user, using not only speech but also nonverbal modalities such as gesture, gaze,
intonation and posture (Cassell, Sullivan et al., 2000). We have constructed several such systems that
use machine vision and speech recognition to sense the user's speech, gesture, body posture and
intonation, and use models of the relationship between verbal and nonverbal behavior to make the
animated computer character respond appropriately. Our research on ECAs has begun to serve the goals
of our research on story listening systems as we have developed the notion of shared reality to describe
systems in which virtual collaborators can share real objects with their human users. This is the case for
MACK, an animated embodied robot in a kiosk who can share a real map with users, and for two recent
story listening systems that we have built:
• GrandChair, a virtual grandchild who elicits life stories from seniors; speech recognition and
sensors in a rocking chair that the user sits in allow the virtual grandchild to give appropriate
feedback, while the system videotapes the stories for later viewing by the senior and his/her
family (Smith, 2000).
• Sam, a virtual peer designed to engage in particular kinds of storytelling play with children,
described further below.
Sam the Castlemate (Cassell, Ananny et al., 2000) was designed to
encourage children to produce decontextualized language and
metacognitive reflection in the context of peer storytelling, by
modeling and scaffolding the child in those activities, and inviting
the child to scaffold and coach in return. Sam is a virtual child who
invites children to participate in collaborative and conversational
storytelling play with real toys. The Sam system has two
components: an embodied conversational agent – a life-size gendernon-specific child named Sam – and a toy castle with several plastic
figurines. Sensors, audio threshold, and floor mats allow Sam to
recognize when the child is speaking, what room the child is currently playing in, and which figurine the
child is using. Sam is projected on a screen behind the castle, and can both tell stories, using a recorded
child’s voice, and listen to the real child’s stories, responding with appropriate feedback (such as “cool,”
Story Listening Systems - 16 -
“then what happened?” and short comments (such as “I’ll put the toy in the magic tower so you can take
a turn”). The physical toy castle that Sam and the real child share extends into the virtual world. The
figurines can exist in either the physical world or on the screen, but never both, and Sam and the child
can pass them back and forth between their worlds by way of the magic tower. Sam has a database of
rather short stories that involve one or two characters played out by the figurines, and involve several
locations, which Sam illustrates by moving the figurines among the different rooms of the castle. All of
the stories are told in the first person, begin with the introduction of a scene, and involve complicating
actions (e.g., losing a horse) and resolution (e.g., found the horse), as well as appropriate introduction of
referents, and use of cohesive devices. The stories were written by several members of the Sam research
team, and then re-written for us by a local nine year old who then recorded the stories in her own voice
(and with her own sound effects). In a newer version of Sam (post- the evaluation reported below), all
stories also demonstrate third person narrative voice and embedded reported speech (“character voice”).
A recent empirical study examines the effect of Sam on
collaborative peer-like and metalinguistic behavior, and use of
decontextualized language. This study employed a 2x2 design,
comparing (a) single children to pairs of children, and (b)
interacting with Sam vs. interacting with the castle but without
Sam. The single child vs. dyad condition allows us to examine
the effect of Sam on collaborative behaviors among real peers,
as well as to get a baseline of peer-peer collaborative
storytelling play. All children came from a mixed ethnicity,
mixed socio-economic status public school in one of the towns
surrounding Boston. Only girls participated in this particular
study, as the version of Sam that they played with employed a pink castle, and some boys refused to
participate for that reason (our current version of Sam plays with a wooden castle, and this problem has
been alleviated). All children were aged between 4 years 6 months and 5 years 6 months. 31 children
have been videotaped interacting with the castle, in pairs and alone, with and without the Sam virtual
peer. Experimental procedure was identical for all conditions: the castle and figurines, whether or not
Sam was present, were introduced in the following way: “This toy is good for telling stories. See the
castle? And the little toy? You can use these to tell stories. Ok, do you want me to show you? I'll tell a
story with the toy, and then you can tell a story.” In the Sam condition, the experimenter added “That's
Sam. Sam likes to tell stories and listen to your stories. Sam knows that it is my turn to tell the story
because I will take the toy from the magic tower.” The experimenter finishes the instructions by saying,
“Tell as many stories as you like [in Sam condition: “with Sam”] and then come and get me.” For the
remainder of the session, the experimenter remained outside the room, while a video camera recorded the
child.
Sam involves all four traits of the model described above: it (i) engages children’s oral storytelling skills,
(ii) involves peers as playmates both in the system and with the system, (iii) invites whole-body play with
familiar toys, and (iv) allows children to produce their own personally meaningful content. For this
reason, I spend some time below analyzing children’s use of the Sam system. In what follows, I first
describe some general characteristics of children’s play with the system, and then use transcripts of
children’s interaction with Sam to give examples of children’s performance on the literacy predictors
outlined above.
Children’s play with Sam is striking in its naturalness. The children enjoy playing with Sam, and appear
to enter into the game of playing with a virtual peer, perhaps from previous play with puppets, or from
experiences with imaginary playmates. The children do know that Sam is a toy; as one child remarked:
Story Listening Systems - 17 -
“[Sam’s] more complicated because like you have toys that like do only
like one thing -- Sam does a lot of things. He tells different stories - not that many. It's way complicated”.
On the other hand, they are still concerned that Sam listen to them, as demonstrated in the following
interaction between a child and the experimenter. In this instance the system incorrectly sensed that the
child had left, and so Sam went through the farewell part of the interaction, which led the child to fetch
the experimenter from her observation post outside the experimental room:
CHILD:
Sam said bye. He said bye to me and left.
ADULT:
Uh oh. I wonder what that means.
CHILD:
Sam’s gone
ADULT:
Sam’s gone.
Let me see if I can fix it.
happens with computers. Sometimes things break.
CHILD:
But, he wasn’t listening? Was Sam listening?
That’s what
The children who played with Sam, either by themselves, or with another child, told a number of
complete and fairly long stories. This is in contrast to children playing by themselves in the absence of
the Sam virtual peer, who may very well be telling themselves silent stories, but who do not vocalize
those stories. These children become bored quite quickly, and start to take the castle apart (also a
problem when two children played with each other, without Sam, as can be seen below).
But a complicated, even an open-ended, toy does not suffice in literacy development. And, if children
playing alone don’t engage in literacy practice, then one might ask: why not simply provide children with
peers, since peers have been demonstrated to be important in emergent literacy? As a baseline story
between peers, and in partial response to that question, below is an example of two children playing with
the castle (and without Sam).
CHILD #1: It hangs there -CHILD #2: Do you want me to do it?
CHILD #1: No, I can.
CHILD #2: She said nothing's going to move for it.
CHILD #1: But I'm just scared,
CHILD #2: I am, too. Let's turn on the lights. Now you won't be
scared.
CHILD #1: Because ghosts do not like light, right?
CHILD #2: No.
CHILD #1: You broke this after I had fixed it.
CHILD #2: Not me.
CHILD #1: It's probably the ghost.
CHILD #2: There's no such thing as monsters. Did that door just open,
or was it just my imagination?
CHILD #1: It was just your imagination.
CHILD #2: No. I think it was just the wind. I'm having nightmares.
CHILD #1: Me, too.
CHILD #2: I want to sleep. I want to sleep. I hope I am.
In this interaction there is a seamless transition between off-topic talk (as the children broke, and
attempted to fix, the chandelier in the toy castle) and first-person narrative talk as the two children played
at being friends in a haunted house. Knowing which segment of talk fulfills which function requires
access to the physical context, as the children are not using explicit contextualizing cues. This kind of
conversation is not a bad thing, as it’s perfectly comprehensible to all participants. On the other hand, it
isn’t demonstrating the kinds of decontextualizing storytelling behaviors that we know are correlated
with literacy skills. We can compare the previous interaction to the following instance of one child
playing with Sam:
SAM: Today, I'm gonna ride horses in the meadow! My dad, the king, said
I can ride the, my big horse named Star. Oh no! Star has been stolen.
Story Listening Systems - 18 -
I'd better go tell the sheriff. Oh, sheriff, my favorite horse Star
has been stolen, I don't know where she is. Oh, no. No need to worry.
The kind old lady from the other side of the forest has brought her
back. And she’s just coming back now. Whoopie! Thanks. Come on Star!
CHILD: Once there was a little prince. His mom and dad were really,
really rich. That's what made them the king and the queen. He liked
his life so much. He got a really fancy bedroom with a big window
that was so high up he could see the whole entire land. One day he
really wanted to do something. He went downstairs so he could ask
his mother what there was to do. He said, Mom, what can I do? Then
his father came into the room and they had a big surprise for him.
Well, dear, we got more money so that means you can do whatever you
want to do. Do you have any ideas of what you want to do today?
Well, I want, I've always wanted to have, I've always wanted to have
very, I've always wanted to have a crown. Cause. Well, we can get
you one. Do you have anything else that's so much fun? Well, I
want, today I want to learn how to do Irish step dancing. Cause St.
Patrick's Day is coming up and I want to do some of that to entertain
you. Well, OK. They got him a crown, then they went to a class
where he could learn how to do Irish step dancing. And he learned
everything. And on St. Patrick's Day when he woke up he not walked
down the stairs like you usually do to be careful, he ran, ran and
then he tripped and then he fell and then he started crying. Then
his mom and his dad took him to the hospital and the doctor said he
broke his leg. That meant he had to stay in his bed and he couldn't
do Irish step dancing. So he was really sad. But then he got an
idea. Probably he could show his mom and dad how to do Irish step
dancing. Like draw all the steps and how he did it. Because he had
a magic potion, so he put spells on his pictures and the pictures
showed his mom and dad how to do Irish step dancing. Then he went
upstairs and went into bed and closed the door. The End.
Note that this child’s story continues Sam’s theme of a king and his child, but then launches into topics
that clearly come from the child’s own life or imagination (Irish step dancing). Many of the children who
worked with Sam continued themes from the Sam story that they heard. In addition, this child’s story is
well constructed, uses third person narrator voice and embedded reported speech for several different
characters, and includes evaluative comments (“so he was really sad”). In fact, the majority of children
playing with Sam told stories in the third person (despite the fact that Sam didn’t model that), while the
majority of children playing with another child, without Sam, told stories in the first person, reflecting
our earlier findings with the StoryMat system. And, those children who did tell first person stories in the
single-child-Sam condition appeared to be explicitly continuing one of Sam’s stories, as in the following
example:
SAM: OK, my turn. I'm going to have a party and I'm going to invite
whoever I want. My mom, the queen said I could. Oh look! Here
Chris comes. I'm going downstairs to meet him. “Hi Chris, you're the
first one here. What do you want to do?” “Well, we can go ask the
mirror, who is going to come to your party?” “Oh yeah. Oh, mirror,
who is going to come to my party?” “Oh, people from all the way over
the land.” “Oh yeah. Here, I see Cheryl, Brad, and Sean. Oh, this
is going to be a great party.” I'll put the toy in the magic tower so
you can tell a story.
CHILD: La, la, la, la. Oh, it's my birthday today, and mommy and daddy
are going to be out. I better get setted up. Da, da, da, da, da.
Dee, dee, dee, dee. Uh oh.
SAM: And then what happened?
CHILD: I see my friends coming. I need to set up the party still. Oh
no. Hi friends. Where's the party. Oh, I haven't set it up yet.
Would you help me? Sure. I'll watch. Now tell me, my friends are
Story Listening Systems - 19 -
they coming or not. Da, da, da, da. The yummy dessert is on the
table. Where's my [XXX]. OK. Coming up. See any friends yet?
Yup. I see Carla, Meg and Sheg. That's everyone I invited and you.
Oh. This is going to be a great party. What's with the present? Oh
no. I forgot it. I can run home and get it. I'll let the others
in. Oh hi, friends. Yeah, I know I'm late. Where's the [. . .]
Continuing themes from Sam’s story is not the only kind of collaboration that was apparent in the
interactions. Single children seem to engage in peer-like collaborative interaction with Sam, talking as
they might with another child, in particular as they might with a slightly younger child:
CHILD: You done, Sam? OK.
SAM: I'll put the toy in the magic tower so you can tell a story.
CHILD: What should I tell, Sam? Do you have an idea? Hmmmm.
SAM: tell me a story
CHILD: Oh, the girl was happy. She came back from- her husband was
there, she was very happy.
Everyone, I mean everyone knew she
was a good girl. She always had fun playing with her sisters.
CHILD: You’re silly, Sam!
SAM: I’ll put the toy in the magic tower and you can tell a story.
CHILD: Hey Sam, want to talk about like where are we from? OK?
Sam, I’ll tell a story that’s really cool, about where I’m from.
SAM: Cool.
CHILD: Once there was a girl, her name was Simone. That’s me. She
went to the house, she loved it, and then she married a boy named
Sam. Oh, sorry Sam, I messed up.
In fact, children ask Sam to continue their stories, and even coach Sam in storytelling, much as if they
were engaging in reciprocal instruction (Palincsar and Brown, 1984), and in line with Stone and
Christie’s description of peer interaction around literacy activities (Stone and Christie, 1996).
SAM: Cool. OK, my turn. One day me and my friend were playing around
and we heard this loud cry [. . .] I’ll put the toy in the magic
tower so you can tell a story.
CHILD:
[talking to Sam] Try to make a longer story next time.
It's
like this. The little boy was outside. He flipped all around and he
went inside, he did a flip [. . .]
Children’s behavior in the dyad-with Sam condition allows us to examine the effect of Sam on children’s
prosocial collaborative behaviors with one another. And, in fact, we do see more prosocial behavior than
when two children play without Sam, and we find more instances of explicit storytelling. Both effects
are illustrated in the following example:
Once upon a time there was a girl named -CHILD #1:
What about Anna?
CHILD #2:
And the mom and dad went to a store. She could do whatever
CHILD #1:
she wanted. So, she built everything. She went down the stairs, and
then she broke the lamp. And then she messed everything up. And she
jumped and jumped all over the place. And then she looked in the
magic mirror and said, when is my mother and father coming home? And
the magic mirror said, they're coming home right now. And she ran
upstairs.
Wait a minute,
CHILD #2:
CHILD #1. We can go upstairs.
And she ran upstairs. And she ran upstairs again. So, they
CHILD #1:
didn't find her. And then they were surprised that it was all messed
up. And they didn't even know who it was from. So, then, she came
back down. And they said, Annabelle. Did you do this? And she
said, no. And she was lying.
So, her nose went big?
CHILD #2:
So, then, the mother and father put her bed.
CHILD #1:
Story Listening Systems - 20 -
CHILD #2:
CHILD #1:
that.
CHILD #2:
CHILD #1:
puts the
Because she lied?
Because she lied, and because she wasn't supposed to do
OK. My turn.
Sammy - I want Sammy to do it. I'll put it back. (CHILD #1
toy in the magic tower for Sam to take her turn)
In sum, children modeled their stories after Sam’s, but also told stories as explicit models for Sam to
follow. In doing so, children engaged in decontextualized language use, and metalinguistic comments
about storytelling. Children engaged in prosocial constructive collaboration with Sam, and more often
engaged in those types of behaviors with other children when in Sam’s presence. In this sense, Sam is
acting much like the Vygotskian more capable peer (Vygotsky, 1978), seeming to push children to act at
the top of their individual abilities, or in their zone of proximal development, through the nature of their
social interactions. It is not coincidental, of course, that social interaction be so important to the
acquisition of literacy. As discussed above, in Cameron and Wang’s terms, the outside-in component of
literacy requires “recontextualizing” one’s thoughts so that they make sense to another (Cameron and
Wang, 1999). Talking to Sam, like talking to a slightly older child, or a stranger, appears to require more
top-down structure and reflection, and to push children to use grounding devices (devices to ensure and
establish common ground) in their talk. Children’s interaction with Sam demonstrates that social
interaction, even with a virtual peer, can support individual development in the domain of language use
and collaborative behavior.
Despite these positive, and literacy-supporting, aspects of children’s collaboration with Sam, there are
still serious limitations to the system, and many puzzles. Sam currently carries out no natural language
understanding and thus does not really understand what the child is saying. This leads to occasional
failures in communication as children begin stories and expect Sam to continue them, or as children ask
Sam explicit questions and receive no response. For this reason we are in the process of investigating
options for keyword spotting – understanding single words in the stream of the child’s speech – so that
Sam will have a better idea of what the child is talking about. In terms of tantalizing puzzles, although
the children appear to attend to Sam’s stories – continue the themes, ask for Sam’s advice, coach Sam –
some features of their interaction are peculiarly un-peerlike. For example, many of the children do not
look at Sam when telling their story, or give feedback while Sam is speaking. Finally, a number of the
children who interact with Sam tell stories that are clearly based on interaction with objects in the castle
rather than any top-down story structure (“he went up the stairs, then he swung from the chandelier, then
he went down the stairs, then he went through the door . . .”). These features are all the more fascinating
for being not quite peer-like responses. They are leading us to new experiments where we vary some
aspects of Sam’s performance – embodied or verbal feedback, kinds of story structure, and so forth.
5. Future Work
The burgeoning field of emergent literacy has acknowledged the role of peers, of play, and of narrative in
later academic competence. When technologies have been designed to support literacy, however, these
findings have not been incorporated. The work presented here demonstrates that, first of all, these
findings can be incorporated into technology, and second, when they are incorporated, that children
perform well on tasks that are good predictors of later literacy.
In the research reported here, some striking correlations have turned up between the use of the Story
Listening Systems and three crucial predictors of literacy. Children using SLS were shown to use more
metalinguistic terms, more decontextualized language, and to engage in more pro-social collaboration
with peers. This research has not targeted, however, correlations between the use of SLS and actual
writing literacy. Nor has the work reported here targeted in any systematic way the issue of broadening
the definition of emergent literacy. Both of these issues constitute the goals of my future work. In order
Story Listening Systems - 21 -
to analyze the effects of SLS on actual writing literacy, and to examine use of SLS by children engaging
in a broader spectrum of pre-school language behaviors, we have implemented another Story Listening
System, and are preparing to employ a different methodology, suited to more longitudinal, and more
school-based studies.
5.1. Animal Blocks
Animal Blocks explores an explicit connection between
children’s oral storytelling play with physical toys and their
writing. Animal Blocks resembles an empty storybook, with a
set of small wooden animal figurines lying next to it and a set of
landscape props (a meadow, a lake, a forest). When a single
child or several children choose from the collection of small
wooden animal blocks and landscape objects to play with on the
special platform, a colorful projection of the toys they have
chosen migrates up onto the blank page of the storybook sitting
next to the blocks. The child’s choice determines which
beginning of a written story is projected onto the left-hand side of the white pages. The right-hand side is
blank, except for the words “and then.” A voice prompt (in the voice of a child), invites the child to tell a
story. After the child responds, the voice prompt gives the child the choice of listening to his/her story,
looking at other children’s stories, or typing in a continuation of the story on the page. The child can
continue the story by continuing to play with the physical animal blocks, and by entering text onto each
page (by voice-recording or by using a keyboard, although the rest of the interface in no way resembles a
computer). If the child pages backwards in the storybook s/he finds stories by other children who have
used the interface. As the child pages forward, the current child’s entire story is regrouped on the lefthand side of the page, while the right-hand side displays a prompt to continue. In a sense, Animal Blocks
is an extension of TellTale into the written mode, because the prompting and the page structure during
the child’s storytelling mirrors the segmentation of a story. The hypothesis is that this configuration will
allow children to use decontextualized language even in the presence of the toys they are speaking of,
and that the page structure of the book will encourage overall structure to be applied to the story.
5.2. Researching School-based Literacy
In the school-based context, and even with respect to actual writing activities, I am still interested in the
same characteristics of language, and behavior around language, only as manifested in writing, and
behavior around writing. Thus: (i) decontextualized language including appropriate use of pronouns vs.
full noun phrases, narrative perspective (as in the StoryMat study, first vs. third person narrating of a
character’s words), and telling of stories with an overarching structure vs. associations of disconnected
events; (ii) metalinguistic reflection including explicit talk about storytelling, such as found in the
example above where a child told Sam how to tell a good story, as well as children’s use of different
narrative voices; (iii) prosocial constructive collaboration, such as that demonstrated above where
children integrate words and themes from the SLS stories into their own, and also those instances where
dyads of children playing with the SLS engage in turn-taking with one another (as opposed to trying to
take the toy away from one another, or independent parallel play); and finally (iv) broadened definitions
of emergent literacy, which in this instance means looking at the relationship between different kinds of
discourse behaviors and later writing by building a database of stories for Sam and for Animal Blocks
that includes different kinds of starter stories, such as topic-associating narratives (Gee, 1985; Heath,
1986), rapping stories (Pinkard, in press), and others, and looking at whether children who come from the
cultures where these story types originate are more influenced in later writing literacy and whether
children from other cultures broaden the kinds of stories they are able to tell, and whether this has a
positive influence on their writing skills.
Story Listening Systems - 22 -
This research requires quite a different methodology than the non-school, play-based environments, and
relatively ethnographic approach used previously. Particularly appropriate is the design experiment
(Brown, 1992) or design study (Linn and Hsi, 2000) approach to testing the impact of the technologies on
children’s literacy development. This methodology emphasizes careful observation of learning in a
particular (designed) testing environment for the purposes of redesigning it to improve learning processes
and outcomes. Thus, for example, I plan to set up Sam and Animal Blocks and a paper-and-pencil
writing exercise at three different “activity centers” in classrooms, and allow children to choose to play
with whichever system tempts them most. After a certain period of time (for instance, 8 or 10 weeks),
learning outcomes are examined, and changes are made to the technologies. In this work the focus will
continue to be the 4 to 8 age group (pre-school, kindergarten, first, second grade). Recent discussions
pertaining to predictors of children’s school success and readiness have highlighted the need to begin
intervention before school age (Powell, 1985; O'Neill and Pearce, in preparation), and it is therefore
important to collect before and after statistics from the age group during which the most change is
occurring.
Of course, in school-based studies such as these, children’s initial developmental level is key – that is,
where do the children start before they begin to use the technology? Important to this kind of assessment
is a test that has been standardized on children of the right age group, and that demonstrates good
reliability and internal consistency (such as the Test of Early Language Development – 2nd Edition
(Hresko, Reid et al., 1991). And assessing children’s subsequent level – after the SLS intervention – is
equally important – and not self-evident. Some have used achievement tests, such as the written
expression subtest of the Peabody Individualized Achievement Test – Revised (PIAT-R (Markwardt,
1998)). But researchers in the field of emergent literacy have stressed the need to also examine less
quantifiable features, such as metalinguistic reflection, and expressions that have to do with narrative
expression per se. And, pro-social behaviors can’t be measured from text at all, and instead come from
analyses of children writing stories together, or from assessing children’s knowledge of positive
interaction strategies by having them judge those behaviors when acted out by puppets.
6. Implications and Conclusion
The kinds of learning environments enabled by Story Listening Systems provide opportunities for
children to bridge informal and formal contexts by sanctioning and encouraging modes of play as
avenues for the development of "school" language. But as well as informing the design of contemporary
learning environments, the research reported here also impacts the fields of language acquisition,
cognitive development, and early education. In fact, a recent study has shown that children’s ability to
take multiple perspectives in storytelling is positively correlated with their mathematical skills (O'Neill
and Pearce, in preparation), meaning that development in metalinguistic awareness may have an effect
beyond writing literacy. Intelligent Tutoring Systems might also profit from integrating models of peer
relations of the kinds described here. Tutoring systems that could engage in fluid negotiation of roles –
sometimes the more capable partner, sometimes needing some coaching or scaffolding of its own –
would undoubtedly be engaging to children, but might also lead to strong learning results.
More generally, the systems described here demonstrate a model of collaboration between human and
technology that is quite different than that found in most educational technology. Story Listening
Systems exhibit a kind of collaboration that resembles that found among peers more than the
collaboration (or cooperation) between an expert and a novice. However, unlike the cognitive
dimensions of peer learning companion systems, SLS demonstrate the importance of the social context of
peer collaboration, its playful, spontaneous, personally meaningful dimensions, its ability to evoke a
desire to make oneself understood. All of these dimensions enter into the socio-cultural understanding of
learning, whereby the very fact of monitoring mutual understanding, of watching for how another
understands what one has just said, changes one’s understanding of one’s own words or actions. For
Story Listening Systems - 23 -
literacy learning, this process of appropriation (Rogoff, 1991) is key. Literacy is all about making
meaning for others – about participating in a community of meaning-makers.
Finally, the field of Children’s Media has been concerned with the kinds of media designed for children
and youth. Young audiences have always appropriated content in innovative and provocative ways. But
as makers of new technology, we have the chance to provide children with media that encourage them to
make, to produce, to construct. All of which leads us back to the policy arena and the debate regarding
the pros and cons of technology for children. In the context of this unnecessarily polarized, and often
heated, debate, the proposed research attempts to provide a knowledge base to inform the two sides. In
particular, the current research builds on the assumption that there is no inherent conflict between
technology and physical exploration, play, and the joy of active discovery. Rather, technological tools
can be designed to incorporate the powerful, positive aspects of play and active learning at the same time
as they scaffold and empower more formal language, cognitive and social development.
7. Acknowledgements
Thanks to the many students who have participated in the development of the systems, and the ideas
presented here, in particular Mike Ananny, Linda Lin, Kimiko Ryokai, Elisabeth Sylvan, Catherine
Vaucelle, as well as to the other members of Gesture and Narrative Language group at the MIT Media
Lab. My gratitude to Susan Goldman, Jack Mostow, Laura Saltz, and Barbara Tversky for comments
that improved the manuscript. Thanks also to all of the many children who have participated in the
studies.
8. References
Ananny, M. and J. Cassell (2001). "Telling Tales: A new toy for encouraging written literacy through
oral storytelling.". Society for Research in Child Development Biennial meeting, Minneapolis,
MN.
Auwarter, M. (1986). Development of Communicative Skills: The construction of fictional reality in
children's play. Children's Worlds and Children's Languages. Cook-Gumperz, J. (ed.). Berlin,
Mouton de Gruyter.
Benford, S., B. Bederson, et al. (2000). "Designing Storytelling Technologies to Encourage Collaboration
Between Young Children". Proceedings of CHI, ACM Press.
Bers, M. and J. Cassell (1998). “Interactive Storytelling Systems for Children: Using Technology to
Explore Language and Identity.” Journal of Interactive Learning Research 9(2): 183-215.
Bickham, D., J. Wright, et al. (2001). Attention, Comprehension, and the Educational Influences of
Television. Handbook of Children and Media. Singer, D & J. Singer (eds.). New York, Sage
Publications.
Bokus, B. (1992). “Peer Co-Narration: Changes in Structure of Preschoolers' Participation.” Journal of
Narrative and Life History 2: 252-275.
Brophy, S., G. Biswas, et al. (1999). Teachable Agents: Combining Insights from Learning Theory and
Computer Science. Artificial Intelligence in Education. S. P. Lajoie and M. Vivet. Amsterdam,
IOS Press: 21-28.
Brown, A. L. (1992). “Design experiments: Theoretical and methodological challenges in creating
complex interventions in classroom settings.” Journal of the Learning Sciences 2(2): 141-178.
Bruckman, A. (1995). “The MediaMOO Project: Constructionism and Professional Community.”
Convergence 1(1).
Story Listening Systems - 24 -
Bruner, J. (1990). Acts of Meaning. Boston, Harvard University Press.
Cameron, C. A. and M. Wang (1999). “"Frog, where are you?" Children's narrative expression over the
telephone.” Discourse Processes 28(3): 217-236.
Cassell, J., M. Ananny, et al. (2000). "Shared Reality: Physical Collaboration with a Virtual Peer".
Proceedings of CHI 2000, The Hague, The Netherlands, ACM SIGCHI Conference on Human
Factors in Computing Systems (CHI).
Cassell, J. and K. Ryokai (2001). “Making Space for Voice Technologies to Support Children's Fantasy
and Storytelling.” Personal Technologies 5(3): 203-224.
Cassell, J., J. Sullivan, et al. (2000). Embodied Conversational Agents. Cambridge, MIT Press.
Cazden, C. B. (1976). Play with language and metalinguistic awareness: One dimension of language
experience. Play: Its role in development and evolution. J. Bruner, J. Jolly and K. Sylva. New
York, Basic Books: 603-608.
Chan, T. W. and A. Baskin (1988). "Studying with the Prince: The Computer as a Learning Companion".
Proceedings of International Conference on Intelligent Tutoring Systems, 194 -200, Montreal,
Canada.
Cognition and Technology Group at Vanderbilt (1998). Designing environments to reveal, support, and
expand our children's potentials. A survey of research approaches. S. Soraci and W. J. McIlvane.
Stamford, CT, Ablex Publishing Corporation.: 313-351.
Committee on Information Technology Literacy (1999). Being Fluent with Information Technology.
Washington, D.C., National Research Council.
Cordes, C. M., Edward (2000). Fool's Gold: A Critical Look at Computers in Childhood. Washington,
DC, Alliance for Childhood.
Corsaro, W. (1992). “Interpretive reproduction in children's peer cultures.” Social Psychological
Quarterly 55: 160-177.
Daiute, C., C. H. Campbell, et al. (1993). Young Authors' Interactions with Peers and a Teacher: Toward
a Developmentally Sensitive Sociocultural Literacy Theory. The Development of Literacy
through Social Interaction. C. Daiute. San Francisco, Jossey-Bass Publishers: 41-63.
Damon, W. and E. Phelps (1989). Strategic Uses of Peer Learning in Children's Education. Peer
Relationships in Child Development. T. Berndt and G. Ladd. New York, Wiley: 135-157.
Dehn, D. M. and S. v. Mulken (1999). The Impact of Animated Interface Agents: A Review of Empirical
Research. M.S. Saarbrucken, Germany, University of Saarland.
Dickinson, D. K. and P. O. Tabors (1991). “Early literacy: linkages between home, school and literacy
achievement at age five.” Journal of Research in Childhood Education 6(1): 30-46.
Doise, W. (1990). The development of individual competencies through social interaction. Children
helping children. H. C. Foot, M. J. Morgan and R. H. Shute. Chichester, J. Wiley & sons: 43-64.
Druin, A., J. Stewart, et al. (1997). "KidPad: A design collaboration between children, technologists, and
educators". Proceedings of CHI '97, Atlanta, GA, ACM Press.
Dyson, A. H. (1993). A sociocultural perspective on symbolic development in primary grade classrooms.
The development of literacy through social interaction. C. Daiute. San Francisco, CA, JosseyBass Publishers.
Story Listening Systems - 25 -
Feagans, L. and M. Appelbaum (1986). “Validation of language subtypes in learning disabled children.”
Journal of Experiment Psychology 78: 358-364.
Gallas, K. (1992). “When the children take the chair: A study of sharing time in a primary classroom.”
Language Arts 69: 172-182.
Gee, J. (1986). “Units in the Production of Narrative Discourse.” Discourse Processes 9: 391-422.
Gee, J. P. (1985). “Two styles of narrative construction and their linguistic and educational
implications.” Journal of Education 171(1): 97-115.
Glos, J. and J. Cassell (1997). "Rosebud: Technological Toys for Storytelling". Proceedings of CHI '97,
Atlanta, GA, ACM Press.
Gombert, J. E. (1992). Metalinguistic Development. Chicago, IL, The University of Chicago Press.
Goncu, A. (1993). “Development of Intersubjectivity in the Dyadic Play of Preschoolers.” Early
Childhood Research Quarterly 8: 99-116.
Hall, N. (2000). Literacy, Play, and Authentic Experience. Play and Literacy in Early Childhood. K.
Roskos and J. Christie. Mahwah, NJ, Lawrence Erlbaum: 189-204.
Hayes-Roth, B. and R. v. Gent (1997). "Story-Making with Improvisational Puppets". Proceedings of
Autonomous Agents 97, Marina Del Rey, CA.
Heath, S. B. (1983). Ways with words: Language, life, and work in communities and classrooms.
Cambridge, MA, Cambridge University Press.
Heath, S. B. (1986). Sociocultural contexts of language development. Beyond language: social and
cultural factors in schooling language minority students. Los Angeles, CA, Evaluation,
Dissemination and Assessment Center: 143-186..
Heath, S. B. (1986). What no bedtime story means: Narrative Skills at home and school. Language
Socialization Across Cultures. B. B. Schieffelin and E. Ochs. Cambridge, Cambridge, MA: 97124.
Hiebert, E. H. and T. E. Raphael (1998). Early literacy instruction. Fort Worth, TX, Harcourt Brace
College Publishers.
Hinkle, D. (2000). School Involvement in Early Childhood, National Institute on Early Childhood
Development and Education, U.S. Department, Office of Educational Research and
Improvement.
Hresko, W. F., D. K. Reid, et al. (1991). Test of early language development. Austin, TX, Pro-Ed.
Johnson, W. L., J. W. Rickel, et al. (2000). “Animated Pedagogical Agents: Face-to-Face Interaction in
Interactive Learning Environments.” International Journal of Artificial Intelligence in Education
11: 47-78.
Katsarou, E. (1992). Home Culture and School Ethos: Four Students' Second Language Literacy and
Academic Achievement (Sociocultural Environment). Ph.D. thesis in Education. Chicago,
University of Illinois at Chicago.
Kim, M. Y. (1995). "Creative Multimedia for Children: Isis Story Builder". Proceedings of CHI '95,
Denver, Colorado, ACM.
Labov, W. (1972). The Logic of Nonstandard English. Language in the Inner City: Studies in the Black
English Vernacular. Philadelphia, PA, University of Pennsylvania Press: 201-240.
Story Listening Systems - 26 -
Lee, C. D. (1992). “Literacy, cultural diversity and instruction.” Education and Urban Society 24: 279–
291.
Linn, M. C. and S. Hsi (2000). Computers, teachers, peers: Science learning partners. Mahwah, NJ,
Lawrence Erlbaum Associates, Inc.
Markwardt, F. C. (1998). Peabody Individual Achievement Test - Revised. Circle Pines, MN, Academic
Guidance Service.
Michaels, S. (1981). “Sharing Time: Children's Narrative Styles and Differential Access to Literacy.”
Language in Society 10: 423-442.
Michaels, S. (1986). Narrative presentations: An oral preparation for literacy with first graders. The
Social Construction of Literacy. J. Cook-Gumperz. Cambridge, CUP: 94-116.
Miller, P. and L. Hoogstra (1992). Language as Tool in the Socialization and Apprehension of Cultural
Meanings. New Directions in Psychological Anthropology: Appraisal and Prospects. T.
Schwartz, G. M. White and C. Lutz. Cambridge, Cambridge University Press: 83-101.
Moreno, R., R. E. Mayer, et al. (2000). "Life-like pedagogical agents in constructivist environments:
cognitive consequences of their interaction". Proceedings of ED-MEDIA, Charlottesville, VA,
AACE.
Mostow, J., G. S. Aist, et al. (in press). 4-Month Evaluation of a Learner-controlled Reading Tutor that
Listens. Speech Technology for Language Learning. P. DeCloque and M. Holland. The
Netherlands, Swets & Zeitlinger Publishers.
Nelson, K. (1989). Narratives from the Crib. Cambridge, MA, Harvard University Press.
Nelson, K. (1996). Language in Cognitive Development. New York, Cambridge University Press.
Neuman, S. B. and K. Roskos (1992). “Literacy Objects as Cultural Tools: Effects on Children's
Literacy Behaviors in Play.” Reading Research Quarterly 27: 202-225.
Nicolopoulou, A. (1996). Narrative Development in Social Context. Social Interaction, Social Context,
and Language: Essays in Honor of Susan Ervin-Tripp. J. G. D.I. Slobin, A. Kyratzis, and J. Guo.
Mahwah, N.J., Lawrence Erlbaum Associates: 369-390.
O'Neill, D. K. and M. Pearce (in preparation). "A new perspective on the predictive relation between
narrative perspective-taking ability in preschoolers and academic competence". Department of
Psychology, University of Waterloo.
Palincsar, A. and A. L. Brown (1984). “Reciprocal Teaching of Comprehension-fostering and
Comprehension-monitoring Activities.” Cognition and Instruction 1: 117-175.
Pellegrini, A. and L. Galda (1993). “Ten Years After: A Reexamination of Symbolic Play and Literacy
Research.” Reading Research Quarterly 28: 162-177.
Peterson, C., B. Jesso, et al. (1999). “Encouraging narratives in preschoolers: an intervention study.”
Journal of Child Language 26: 49-67.
Peterson, C. and A. McCabe (1983). Developmental psycholinguistics: Three ways of looking at a child's
narrative. New York, Plenum.
Pinkard, N. (in press). “Lyric Reader: An Architecture for instrinsically motivating and culturally
relevant reading environments.” Interactive Learning Environments.
Powell, D. R. (1985). Enabling Young Children to Succeed in School. Washington, D.C., American
Educational Research Association.
Story Listening Systems - 27 -
Preece, A. (1992). “Collaborators and Critics: The Nature and Effects of Peer Interaction on Children's
Conversational Narratives.” Journal of Narrative and Life History 2(3): 277-292.
Resnick, M., A. Bruckman, et al. (1996). “Pianos Not Stereos: Creating Computational Construction
Kits.” Interactions 3(6).
Resnick, M., F. Martin, et al. (1998). "Digital Manipulatives: New toys to think with". Proceedings of
CHI 98. ACM Press.
Rogoff, B. (1991). Social interaction as apprenticeship in thinking: guided participation in spatial
planning. Perspectives on Socially Shared Cognition. L. Resnick, J. Levine and S. Teasley: 349364.
Scarlett, W. G. and D. Wolf (1979). “When It's Only Make-Believe: The Construction of a Boundary
Between Fantasy and Reality in Storytellling.” New Directions for Child Development 6: 29-40.
Shatz, M. and R. Gelman (1973). “The development of communication skills: Modifications in the
speech of children as a function of listener.” Monographs of the Society for Research in Child
Development. 38(5).
Short, J., E. Williams, et al. (1976). The Social Psychology of Telecommunications. New York, Wiley.
Singer, J. L. (2001) Learning Through Play. http://www.members.home.net/learnthruplay/. Accessed on
January, 2001.
Smith, J. (2000). GrandChair: Conversational Collection of Family Stories. MA thesis in Media Arts and
Sciences. Cambridge, MA, MIT.
Snow, C. E. (1983). “Literacy and language: Relationships during the preschool years.” Harvard
Educational Review 53(2): 165-187.
Snyder, T. and J. Wirt (1998). The Condition of Education 1998 NCES Report Number 98013
http://nces.ed.gov/pubs98/condition98/index.html, National Center for Education Statistics.
Steiner, K. E. and T. G. Moher (1992). "Graphic Story Writer: An Interactive Environment for Emergent
Storytelling". Proceedings of CHI '92, Monterey, California, ACM Press.
Stone, S. J. and J. F. Christie (1996). “Collaborative Literacy Learning during Sociodramatic Play in a
Multiage (K-2) Primary Classroom.” Journal of Research in Childhood Education 10(2): 123-33.
Sulzby, E. (1996). Roles of oral and written language as children approach conventional literacy.
Children's early text construction. C. Pontecorvo et al.. Mahwah, NJ, Lawrence Erlbaum: 25-46.
Taylor, M. (1999). Imaginary Companions and the Children Who Create Them. Oxford, Oxford U.
Press.
Teale, W. H. and E. Sulzby (1986). Emergent literacy as a perspective for examining how young chidren
becmoe writers and readers. Emergent Literacy: writing and reading. W. S. Teale, E. Norwood,
NJ, Ablex: vii-xxv.
Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes.
Cambridge, MA, Harvard University Press.
Wartella, E. and N. Jennings (2001). “Children and Computers: New Technology--Old Concerns.” The
Future of Children 10(2): 31-43.
Wartella, E., B. O'Keefe, et al. (2000). Growing up with Interactive Media. New York, Markle
Foundation.
Story Listening Systems - 28 -
Wells, G., Ed. (1981). Learning Through Interaction: The Study of Language Development. Language at
Home and at School. Cambridge, Cambridge University Press.
Whitehurst, G. J., F. Falco, et al. (1988). “Accelerating Language Development through Picture-book
Reading.” Developmental Psychology 24: 552-558.
Whitehurst, G. J. and C. J. Lonigan (1998). “Child development and emergent literacy.” Child
Development 69(3): 848-872.